Overview
- At the AGI-Next summit organized by Tsinghua’s Beijing Key Lab and Zhipu AI on January 10–11, leading researchers and executives outlined how China’s AI strategy is shifting from chat interfaces to task‑oriented agents.
- Tencent chief scientist Yao Shunyu said consumer use feels little changed, but enterprise deployment is accelerating as companies prefer the strongest models where greater intelligence yields higher productivity and earnings.
- Yao contrasted flagship examples by market, citing ChatGPT as consumer‑facing and Claude Code as emblematic of fast‑moving enterprise workflows that rely on natural‑language programming.
- Zhipu founder Tang Jie credited DeepSeek’s early‑2025 debut with catalyzing the 'AI does tasks' paradigm and said Zhipu integrated coding, reasoning and agent abilities in GLM‑4.5, with the recent GLM‑4.7 claiming large gains in Agent and Coding capabilities.
- Qwen technical lead Lin Junyang argued that the model itself is the product and described agents evolving toward long‑duration, autonomous work with prospects for embodied reasoning across virtual and physical environments.